1 Leaf phenology as one important driver of seasonal ...

1

Leaf phenology as one important driver of seasonal changes in isoprene emission in 1

central Amazonia 2

Eliane G. Alves1, Julio Tóta2, Andrew Turnipseed3, Alex B. Guenther4, José Oscar W. 3

Vega Bustillos5, Raoni A. Santana2, Glauber G. Cirino6, Julia V. Tavares1, Aline Lopes1, 4

Bruce W. Nelson1, Rodrigo A. de Souza7, Dasa Gu4, Trissevgeni Stavrakou8, David K. 5

Adams9, Jin Wu10, Scott Saleska11, Antonio O. Manzi12. 61 Department of Environmental Dynamics, National Institute for Amazonian Research 7

(INPA), Av. André Araújo 2936, CEP 69067-375, Manaus-AM, Brazil.82 Institute of Engineering and Geoscience, Federal University of West Para (UFOPA), 9

Rua Vera Paz s/n, CEP 68035-110, Santarem-PA, Brazil. 103 2B Technologies, Inc., 2100 Central Ave., Boulder, CO 80301, U.S.A. 114 Department of Earth System Science, University of California, Irvine, CA 92697, USA. 125 Chemistry and Environment Center, National Institute for Energy?? and Nuclear 13

Research (IPEN), Av. Lineu Prestes 2242, CEP 05508-000, São Paulo-SP, Brazil.146 Department of Meteorology, Geosciences Institute, Federal University of Para, Belém, 15

PA 66075-110, Brazil 167 Meteorology Department, State University of Amazonas (UEA), Av. Darcy Vargas 17

1200, CEP 69050-020, Manaus-AM, Brazil. 188 Royal Belgian Institute for Space Aeronomy, Avenue Circulaire 3, 1180, Brussels, 19

Belgium. 209 Centro de Ciencias de la Atmósfera, Universidad Nacional Autónoma de México, Av. 21

Universidad 3000, 04510, Mexico city, Federal District, Mexico. 22

Biogeosciences Discuss., https://doi.org/10.5194/bg-2018-26Manuscript under review for journal BiogeosciencesDiscussion started: 6 March 2018c© Author(s) 2018. CC BY 4.0 License.

2

10 Department of Environmental and Climate Sciences, Brookhaven National Laboratory, 23

Upton, NY 11973, USA. 2411 Ecology and Evolutionary Biology Department, University of Arizona, Cherry Avenue 25

and University Boulevard, Tucson, AZ 85721, USA. 2612 National Institute for Spatial Research, Center of Weather Forecasting and Climate 27

Studies, Rod. Presidente Dutra, km 40, Cachoeira Paulista-SP, Brazil. 28

corresponding author: +55 92 3643 3238; [email protected]

30

Abstract 31

Isoprene fluxes vary seasonally with changes in environmental factors (e.g., solar 32

radiation and temperature) and biological factors (e.g., leaf phenology). However, our 33

understanding of seasonal patterns of isoprene fluxes and associated mechanistic controls 34

are still limited, especially in Amazonian evergreen forests. In this paper, we aim to 35

connect intensive, field-based measurements of canopy isoprene flux over a central 36

Amazonian evergreen forest with meteorological observations and with tower-camera 37

leaf phenology to improve understanding of patterns and causes of isoprene flux 38

seasonality. Our results demonstrate that the highest isoprene emissions are observed 39

during the dry and dry-to-wet transition seasons, whereas the lowest emissions were 40

found during the wet-to-dry transition season. Our results also indicate that light and 41

temperature can not totally explain the isoprene flux seasonality. Instead, the camera-42

derived leaf area index (LAI) of recently mature leaf-age class (e.g. leaf ages of 3-5 43

months) exhibits the highest correlation with observed isoprene flux seasonality 44

(R2=0.59, p<0.05). Attempting to better represent leaf phenology in the Model of 45


3

Emissions of Gases and Aerosols from Nature (MEGAN 2.1), we improved the leaf age 46

algorithm utilizing results from the camera-derived leaf phenology that provided LAI 47

categorized in three different leaf ages. The model results show that the observations of 48

age-dependent isoprene emission capacity, in conjunction with camera-derived leaf age 49

demography, significantly improved simulations in terms of seasonal variations of 50

isoprene fluxes (R2=0.52, p<0.05). This study highlights the importance of accounting for 51

differences in isoprene emission capacity across canopy leaf age classes and of 52

identifying forest adaptive mechanisms that underlie seasonal variation of isoprene 53

emissions in Amazonia.54

55

1. Introduction 56

Isoprene is considered the dominant contribution to Biogenic Volatile Organic 57

Compound (BVOC) emission from many landscapes and represents the largest input to 58

total global BVOC emission, which has the magnitude of 400-600 Tg C y-1 ( see Table 1 59

of Arneth et al., 2008). This compound regulates large-scale biogeochemical cycles. For 60

example, once in the atmosphere, isoprene has implications for chemical and physical 61

processes due to its reactivity, influences on the atmospheric oxidative capacity, as well 62

as its potential to form secondary organic aerosols (Claeys et al., 2004), which interact 63

with solar radiation and act as effective cloud condensation nuclei. Moreover, carbon 64

dioxide is believed to be the fate of almost half of the carbon released in the form of 65

BVOCs (Goldstein and Galbally, 2007) and, as BVOC emissions are regarded as highly 66

significant for ecosystem productivity (Kesselmeier et al., 2002) with isoprene being the 67

most emitted hydrocarbon, it thereby plays an important role in carbon balance.68


4

Tropical forests are the largest source of isoprene to the atmosphere, contributing 69

almost half of the estimated global annual isoprene emission, according to Model of 70

Emissions of Gases and Aerosols from Nature (MEGAN) estimates (Guenther et al., 71

2006). Given that the Amazon basin is the largest territorial contribution to global 72

tropical forests, this ecosystem is thought to be one of the most important sources of 73

isoprene to the global atmosphere. 74

Recently, remotely sensed observations of multiple years have revealed seasonal 75

changes in isoprene emission over the Amazonian rainforest (Barkley et al., 2008, 2009, 76

2013, Bauwens et al., 2016). Apart from these remotely sensed data, only a few studies 77

based on in situ data exist (Alves et al., 2016; Andreae et al., 2002; Kesselmeier et al., 78

2002; Kuhn et al., 2004b; Yáñez-Serrano et al., 2015). Some of these in situ studies 79

indicate that environmental factors such as solar radiation and temperature are primary 80

drivers of isoprene (Andreae et al., 2002; Kesselmeier et al., 2002; Kuhn et al., 2004b; 81

Yáñez-Serrano et al., 2015).82

Canopy phenology has been suggested as the primary cause of seasonal changes 83

of photosynthesis in Amazonian ecosystems (Wu et al., 2016), a suggestion in agreement 84

with reports on phenology effects at the Amazonian tree species level (Kuhn et al., 85

2004a). Given that photosynthesis provides substrates and energy for isoprene production 86

and that isoprene is not stored within leaves, canopy phenology could therefore be an 87

important seasonal driver in isoprene emissions (Alves et al., 2014, 2016). However, 88

even though these factors – solar radiation, temperature, and leaf phenology – have been 89

noted as important drivers of seasonal isoprene emissions, the way in which they control 90

seasonal emissions remains poorly represented in biogeochemical models. 91


5

In terms of modeling of isoprene emission from Amazonia, when light and 92

temperature are considered, MEGAN is a satisfactory tool for predicting short-term 93

changes in isoprene emissions (Karl et al., 2004, 2007). However, when long-term 94

changes are taking place, other factors, some still unknown, might be acting together, 95

which add uncertainties to isoprene seasonal emission estimates. 96

In this study, we present observations of seasonal variation of isoprene flux, solar 97

radiation, air temperature and canopy phenology from a primary rainforest site in central 98

Amazonia. The questions addressed are: (i) how much can seasonal isoprene fluxes be 99

explained by variations in solar radiation, temperature and leaf phenology, and (ii) how 100

can a consideration of leaf phenology observed in the field help to improve model 101

estimates of seasonal isoprene emissions. To this end, we correlate ground-based 102

isoprene flux measurements with environmental factors (light and temperature) and a 103

biological factor (leaf phenology). We compare seasonal ground-based isoprene flux 104

measurements to OMI satellite-derived isoprene flux. Lastly, we perform two simulations 105

with the MEGAN 2.1 to estimate isoprene fluxes: (1) with standard emission algorithms 106

and (2) with a modification in the leaf age algorithm derived from observed leaf 107

phenology. 108

109

2. Material and methods 110

2.1. Site Description - Cuieiras Biological Reserve – K34 site 111

Isoprene fluxes were measured at the 53 m K34 tower (2°36' 32.6" S, 60° 12' 112

33.4" W) on the Cuieiras Biological Reserve plateau, a primary rainforest reserve 113

approximately 60 km northwest of Manaus in Amazonas state, Brazil (Fig. 1). The K34 114


6

tower has been widely utilized for the past 15 years for a range of meteorological studies, 115

including energy and trace gas fluxes (de Araújo et al., 2010; Artaxo et al., 2013; Tóta et 116

al., 2012) and also tropospheric variables such as precipitable water vapor (Adams et al., 117

2011, 2015). This reserve has an area of about 230 km² and is managed by the National 118

Institute for Amazonian Research (INPA). The site has a maximum altitude of 120 m and 119

the topography is characterized by 31% plateau, 26% slope and 43% valley (Rennó et al., 120

2008). The vegetation in this area is considered mature, terra firme rainforest, and with 121

typical canopy height of 30 m with variation (20-45 m) throughout the reserve. More 122

details about soils and vegetation of this site are provided in Alves et al. (2016). Annual 123

precipitation is about 2500 mm and is dominated by deep atmospheric convection and 124

associated stratiform precipitation, December to May being the wet season and August to 125

September the dry season, when the monthly cumulative precipitation is less than 100 126

mm (Adams et al., 2013; Machado et al., 2004). Average air temperature ranges between 127

24 °C (in April) and 27 °C (in September) (Alves et al., 2016). 128

129

2.2. Isoprene flux – Relaxed Eddy Accumulation system (REA) 130

Isoprene flux measurements were conducted during intensive campaigns of five to 131

six days, during daytime (9:00-16:30, local time), from June 2013 to December 2013 at 132

the K34 tower. The REA system utilized for the isoprene flux measurements was 133

developed by the National Center for Atmospheric Research (NCAR) 134

NCAR/BEACHON REA Cassette Sampler), and has two basic components: 1) the main 135

REA box containing the adsorbent cartridges (stainless steel tubes filled with Tenax TA 136

and Carbograph 5 TD adsorbents) for up/down/neutral reservoirs, microcontroller, 137


7

battery, selection valves, and mass flow controller (200 ml min-1) (MKS Instruments Inc., 138

Model M100B01852CS1BV); and (2) a Sonic Anemometer (RM Young, Model 139

81000VRE) for high-rate wind velocity measurements (10 Hz). This REA system was 140

installed at a height of 48 m on the K34 tower (approximately 20 m above the mean 141

canopy height). 142

The technique segregated the sample flow according to sonic anemometer-derived 143

vertical wind velocity over the flux-averaging period (30 min). Isoprene fluxes (F) from 144

the REA system over this period were estimated from: 145

! = !!!! = !!!(!!" − !!"#$) (1) 146

where b is an empirical proportionality coefficient (described below), !! is the standard 147

deviation of w, and !!" and !!"#$ are isoprene concentration averages in the up and 148

down reservoirs, respectively (Bowling et al., 1998). The b-coefficient was calculated 149

from the sonic temperature and heat flux by re-arranging the same equation, assuming 150

scalar similarity (Monin-Obukhov Similarity Theory):151

! = !!!!!! !!"!!!!"#

(2) 152

The REA sampler was operated with a “deadband” - a range of small !! values, 153

centered on !, over which the air was sampled through the “neutral” line. The deadband 154

used was ±0.6!!. The use of a deadband was advisable, because this increased the 155

differences in the measured concentrations (!!" − !!"#$) by sampling only larger eddies 156

(with larger concentration fluctuations) into the up/down reservoirs, reducing the 157

precision required for the analytical measurements. The b-coefficient was also computed 158

(from Eq. (2)) using the same deadband. For this study, the b-coefficient was calculated 159


8

for every 30 min. flux sampling period. The b-coefficient averaged 0.40 ± 0.06 and the 160

flux measurements were filtered for b-coefficients in the range of 0.3 to 0.6. 161

The air sampling was carried out with two tubing lines for up (+w') and down (-162

w') and one tubing line for neutral sampling air (±0.6!! - deadband), each consisting of 163

approximately 1.5 m long tubes (polytetrafluoroethylene, PTFE) positioned such that 164

they sampled air as close to the sonic anemometer as possible. Each inlet valve at the 165

main REA box prevented air from entering the inactive tube (up- in the case of down 166

sampling (-w') and down - in the case of up sampling (+w'), and both up and down in the 167

case of deadband), which otherwise would compromise the concentration differences 168

between up and down reservoirs and, consequently, the flux calculation. 169

The microcontroller recorded the sonic anemometer data and triggered the 170

segregation valves based on this data. The REA technique requires two initial data points 171

prior to each flux averaging period to be able to segregate the sample flow: (1) a mean 172

vertical wind velocity, ! and (2) !!. The ! determined the direction of the instantaneous 173

vertical wind velocity (!! = ! ! − !) and !! was required to calculate the deadband 174

threshold. Both the value of ! and !! were based on the values obtained from the last 175

flux-averaging period (30 min). The microcontroller stored all the necessary wind and 176

temperature information to compute all the parameters required in the equations (1) and 177

(2). More details on errors and uncertainties of the REA technique are found in section 1 178

(Supplementary Information). 179

1802.3. Isoprene concentrations 181

The isoprene accumulated in the adsorbent cartridges was determined from 182

laboratory analysis. The tube samples were analyzed with a thermal desorption system 183


9

(TD) (Markes International, UK) interfaced with a gas chromatograph/flame ionization 184

detector (GC-FID) (19091J-413 series, Agilent Technologies, USA). After loading a tube 185

in the ULTRA Automatic Sampler (Model Ultra1, Markes International, UK), which was 186

connected to the thermal desorption system, the collected samples were dried by purging 187

for 5 minutes with 50 sccm of ultra-high purity helium (all flow vented out of the split 188

vent) before being transferred (300ºC for 10 min with 50 sccm of ultra-pure nitrogen) to 189

the thermal desorption cold trap held at -10 ºC (Unity Series1, Markes International, UK). 190

During GC injection, the trap was heated to 300°C for 3 min while back flushing with 191

carrier gas (helium) at a flow rate of 6.0 sccm directed into the column (Agilent HP-5 5% 192

Phenyl Methyl Siloxane Capillary 30.0 m X 320 µm X 0.25 µm). The oven ramp 193

temperature was programmed with an initial hold of 6 min at 27 °C followed by an 194

increase to 85 °C at 6 °C min-1 followed by a hold at 200 °C for 6 min. The identification 195

of isoprene from samples was confirmed by comparison of retention time with a solution 196

of an authentic isoprene liquid standard in methanol (10 µg/ml in methanol, Sigma-197

Aldrich, USA). The GC-FID was calibrated to isoprene by injecting 0.0, 23, 35, and 47 198

nL of the gas standard into separate tubes. The gas standard is 99.9% of 500 ppb of 199

isoprene in nitrogen (Apel & Riemer Environmental Inc., USA) and was injected into 200

separate tubes at 11 ml min-1. The calibration curve (0.0, 23, 35, and 47 nL) was made 201

thrice before the analysis of the sample tubes of each campaign, with a mean correlation 202

coefficient equal to R²=0.98. In addition, two standard tubes (with 35 nL of isoprene) 203

were run at every 20 sample tubes to check the system sensitivity. The limit of detection 204

of isoprene was equal to 48.4 ppt. All tube samples were analyzed as described above 205

with the exception of tube samples from June 2013 and July 2013. These were analyzed 206


10

in a TD/GC-MS-FID system from the Atmospheric Chemistry Division, NCAR (see 207

section 1 of supplementary information for more details). 208

Isoprene concentration was determined using the sample volume that was passed 209

through each tube. This volume was measured by integration of the mass flow meter 210

signal and stored within the REA data file. While sampling, the concentration found in 211

the blank tubes connected to the cartridge cassette in the REA box, but without flow, was 212

subtracted from the sample tube concentrations. The resulting concentration was used to 213

calculate isoprene flux (Eq. (1)) in mg m-2 h-1. 214

215

2.4. Tower-camera derived leaf phenology and demography 216

Upper canopy leaf phenology was monitored with Stardot RGB cameras (model 217

Netcam XL 3MP) installed at 53 m height on the K34 tower (Lopes et al., 2016; Wu et 218

al., 2016). The camera monitored forest on well-drained, infertile clay-rich soils of low 219

plateaus. Views were wide-angle and fixed, monitoring the same crowns over time and 220

excluding sky, so that auto-exposure was based only on the forest. Images were 221

automatically logged every two minutes from 09:00h to 12:30h, local time. Only images 222

acquired near local noon and under overcast sky (having even diffuse illumination) were 223

analyzed. Images were selected at six-day intervals. The camera monitored upper crown 224

surfaces of 53 living trees over 24 months (1 December 2011 to 31 November 2013). 225

We used a camera-based tree inventory approach to monitor leaf phenology at this 226

forest site (Lopes et al., 2016; Wu et al., 2016). Specifically, we tracked the temporal 227

trajectory of each tree crown, and assigned them into one of three classes: “leaf flushing” 228

(crowns which showed a large abrupt greening), “leaf abscising” (crowns which showed 229


11

large abrupt greying, which is the color of bare upper canopy branches) or “no change”. 230

We then aggregated our census to the monthly scale to derive the monthly-average 231

percentages of trees with new leaf flushing and with old leaf abscission. The percentage 232

of tree crowns with green leaves (1 – the percentage of tree crowns with leaf abscission) 233

is termed as “green crown fraction” (Wu et al., 2016). We obtained a camera-based 234

canopy LAI by applying the same linear relationship between ground-measured LAI and 235

camera-derived green crown fraction, fitted at another central Amazon evergreen forest, 236

the Tapajós K67 tower site (Wu et al., 2016).237

We also estimated the monthly canopy leaf demography by tracking the post-leaf-238

flush age of each crown's leaf cohort and sorting them into three leaf age classes 239

throughout the year (young: <=2 months; mature: 3-5 months; and old: >=6 months) 240

(Nelson et al., 2014; Wu et al., 2016). By multiplying camera-derived total LAI by the 241

camera-derived fraction of crowns in a given age class, LAIs were derived for the three 242

leaf age classes: young leaf LAI, mature leaf LAI, and old leaf LAI. More details on 243

camera-derived LAI are in section 2 (Supplementary Information).244

245

2.5. Modeled isoprene flux estimates - MEGAN 2.1 246

Isoprene fluxes measured by REA (K34 site) were compared with those estimated 247

by MEGAN 2.1. Isoprene emissions estimated by MEGAN 2.1 account for the main 248

processes driving variations in emissions (Guenther et al., 2012). The isoprene flux 249

activity factor for isoprene (γi) is proportional to emission response to light (γP), 250

temperature (γT), leaf age (γA), soil moisture (γSM), leaf area index (LAI) and CO2 251

inhibition (γCO2) according to Eq. (3):252


12

!! = !!"!"#!!!!!!!!"!!"! (3) 253

where CCE is the canopy environment coefficient. For this study, the canopy environment 254

model of Guenther et al. (2006) was used with a CCE of 0.57. MEGAN 2.1 was run 255

accounting for variations in light, temperature, and LAI. Based on changes in LAI, the 256

model estimated foliage leaf age. Both soil moisture and CO2 inhibition activity factors 257

were set equal to a constant of 1, assuming these parameters do not vary. Details on 258

model settings are found in Guenther et al. (2012). 259

Photosynthetic photon flux density (PPFD) and air temperature inputs for all 260

model simulations were obtained from measurements at K34 tower. PPFD and air 261

temperature measured at tower top, every 30 minutes, were hourly averaged. Data gaps 262

during certain months occurred in 2013, but at least 15 days of hourly average PPFD and 263

air temperature were obtained for model input. LAI inputs were acquired from the 264

Moderate Resolution Imaging Spectroradiometer (MODIS) satellite observations for the 265

same period of the isoprene flux measurements. The level-4 LAI product is composited 266

every 8 days at 1-km resolution on a sinusoidal grid (MCD15A2H) (Myneni, 2015). 267

Additionally, by comparison with the standard MEGAN 2.1 model that uses MODIS-268

derived LAI variation, here we also used LAI fractionated into different leaf ages, which 269

were obtained from tower camera observations (as described in the section above). The 270

number of data inputs to the MEGAN simulations is summarized in table 1. 271

272

2.6. Satellite-derived isoprene flux estimates 273

Top-down isoprene emission estimates over the 0.5 degree region around the tower were 274

obtained by applying a grid-based source inversion scheme (Stavrakou et al., 2009, 2015) 275


13

constrained by satellite formaldehyde (HCHO) columns, measured in the UV-visible by 276

the Ozone Monitoring Instrument (OMI) onboard the Aura satellite launched in 2004. 277

HCHO is a high yield intermediate product in the isoprene degradation process 278

(Stavrakou et al., 2014). The source inversion was performed using the global chemistry-279

transport model IMAGESv2 (Intermediate Model of Annual and Global Evolution of 280

Species) at a resolution of 2º × 2.5º and 40 vertical levels from the surface to the lower 281

stratosphere (Stavrakou et al., 2014, 2015). The a priori isoprene emission inventory was 282

taken from MEGAN-MOHYCAN (Stavrakou et al., 2014, http://emissions.aeronomie.be, 283

Bauwens et al. 2017). Given that the OMI overpass time is in the early afternoon (13:30, 284

local time), and the mostly delayed production of formaldehyde from isoprene oxidation, 285

the top-down emission estimates rely on the ability of MEGAN to simulate the diurnal 286

isoprene emission cycle and on the parameterization of chemical and physical processes 287

affecting isoprene and its degradation products in IMAGESv2. For this study, we use 288

daily (24 hours), mean satellite-derived isoprene emissions derived from January 2005 to 289

December 2013. More details can be found in Stavrakou et al. (2009, 2015) and 290

Bauwens et al. (2016).291

292

3. Results 293

The experimental site of this study showed seasonal variation in air temperature 294

and in photosynthetic active radiation (PAR) (Fig. 2a,b) that was comparable to the 295

seasonality presented by the OMI satellite-derived isoprene fluxes for the K34 site 296

domain (Fig. 2c). The interannual variation in the seasonality of these environmental 297

factors, air temperature and PAR, was correlated to the one presented by the satellite-298


14

derived isoprene fluxes, with the highest correlation found between satellite-derived 299

isoprene fluxes and air temperature. Isoprene fluxes and PAR - R2 ranged from 0.34 to 300

0.83 p<0.05; isoprene fluxes and air temperature - R2 ranged from 0.61 to 0.91, p<0.01, 301

from 2005 to 2013. Maxima and minima of PAR, air temperature, and satellite-derived 302

isoprene fluxes were observed during the dry and the dry-to-wet transition seasons, and 303

the wet and the wet-to-dry transition seasons, respectively. 304

As opposed to the average (2005-2013) flux peaking in September, the 2013 305

results suggest a maximum in October, and are found to be substantially lower during the 306

2013 dry season compared to the average of the dry season estimates (reduction of ~31%) 307

(Fig. 2c). The timing of the maximum is not supported by the ground-based observations, 308

peaking in September, but the magnitude of flux estimates in these two months are in 309

good agreement. In the wet-to-dry transition period, the small reduction in satellite-based 310

isoprene fluxes in July 2013, compared to the neighboring months, is corroborated by a 311

similar behavior in the ground-based isoprene fluxes (Fig. 3d). However, the drop in the 312

observations is much stronger than in the top-down estimates (factor of 3 vs. a 70% 313

difference). 314

Different from satellite-derived fluxes, ground-based isoprene fluxes measured 315

with the REA system have not shown significant correlation with PAR and air 316

temperature for the year 2013 (Table 2 and Fig. 3). Ground-based isoprene fluxes also 317

showed the maximum emission during the dry season (September), but emissions 318

remained high in the beginning of the wet season (December), which was not observed in 319

the seasonal behavior of PAR and air temperature. When averages of air temperature and 320

PAR measured only during the same days of REA isoprene flux measurements were 321


15

compared to isoprene fluxes, the correlations coefficients increased, but were still not 322

statically significant (Table 2).323

The forest leaf quantity, shown as Leaf Area Index (LAI), varied little over the 324

year when the total LAI was examined. However, when total LAI was fractionated into 325

three different leaf age classes – young LAI (<=2 months), mature LAI (3-5 months), and 326

old LAI (>=6 months), seasonal variation of each age class appears (Fig. 4). To 327

understand how those LAI age fractions are related to the isoprene seasonality, ground-328

based fluxes of this compound were compared to the LAI age fractions estimated over the 329

entire year (Fig. 4). The highest emissions were observed when the number of trees with 330

mature leaves (mature LAI) was increasing and the number of trees with old leaves (old 331

LAI) was decreasing. Considering seasonal changes in PAR, air temperature, and mature 332

LAI, the latter presented the highest correlation coefficient, explaining 59% of the 333

seasonal isoprene emission variations (Table 2). 334

Isoprene flux simulations carried out with MEGAN 2.1 reveal similarities with the 335

magnitudes observed during several months. But, MEGAN 2.1 did not fully capture the 336

observed seasonal behavior (Fig. 5). Even though the leaf age algorithm of MEGAN 2.1 337

was parameterized with local leaf phenology observations, giving the highest correlation 338

coefficient with observed fluxes (Table 2), isoprene flux simulations with local 339

CAMERA-LAI inputs showed only a reduction in isoprene flux magnitudes. The 340

seasonal behavior observed was the same as in the estimates from the default MEGAN 341

2.1 with MODIS-LAI inputs. Regressions between averages of observations and 342

MEGAN 2.1 estimates, with CAMERA-LAI and MODIS-LAI inputs, were weak and not 343

statistically significant (Table 2). 344


16

As a sensitivity test, observations of isoprene emission capacity at different leaf 345

ages of a central Amazonian hyper-dominant tree species, Eschweilera coriacea (Alves et 346

al., 2014), were used to parameterize the MEGAN 2.1 leaf age algorithm. Leaf level 347

measurements of isoprene emission capacity are scarce in Amazonia. To the authors’ 348

knowledge, Alves et al. (2014) is the only available data of leaf level isoprene emission 349

capacity at different leaf ages of a central Amazonian tree species, which were therefore 350

used for the MEGAN sensitivity test. 351

Further simulations were performed with modifications in the leaf age emission 352

activity factor (EAF), which is dimensionless and is defined as the emission relative to 353

the emission of mature leaves that are, by definition, set equal to one. A new EAF was 354

assigned for each age class, based on observations of emissions of E. coriacea (Fig. 6). 355

Leaf age fraction distribution was provided with input of LAI from MODIS (MODIS-356

LAI) and from LAI-derived field observations (CAMERA-LAI) (Fig. 4). The simulation 357

with the leaf age algorithm parameterized for EAF changes and with MODIS-LAI was 358

similar to the one without changes in the EAF (MEGAN 2.1 default). The simulation 359

with leaf age algorithm parameterized with changes in the EAF and with CAMERA-LAI 360

inputs showed reduced emissions, but a seasonal curve closer to that of isoprene flux 361

observed at K34 (R2 = 0.52, p<0.05) (Table 2). 362

363

4. Discussion 364

This study addressed two main questions with respect to the seasonality of 365

isoprene fluxes in central Amazonia and identified possible limitations in our current 366

understanding related to these questions. 367


17

4.1. How much can seasonal isoprene fluxes be explained by variations in solar 368

radiation, temperature, and leaf phenology? 369

Our finding that isoprene emissions are higher during the warmer season is 370

consistent with previous findings that emissions from tropical tree species are light 371

dependent and stimulated by high temperatures (Alves et al., 2014; Harley et al., 2004; 372

Jardine et al., 2014; Kuhn et al., 2002, 2004a, 2004b). Indeed, satellite-derived isoprene 373

fluxes (2005-2013 years) were well correlated to PAR and even more to air temperature 374

for all years. However, high ground-based isoprene emissions were observed until late of 375

dry-to-wet transition season, when mean PAR and air temperature were already 376

decreasing. 377

The reasons why satellite-derived isoprene fluxes are weakly correlated to 378

ground-based isoprene fluxes can be attributed to either the difference in the studied 379

scales (e.g. local effects could have major influences on ground-based isoprene fluxes) 380

and/or the uncertainties associated with the methodologies used to estimate or calculate 381

fluxes. The high correlation between satellite-based fluxes and air temperature or PAR is 382

not unexpected, because higher temperatures and solar radiation fluxes favor isoprene 383

emissions. Note however that the satellite-derived fluxes might also be subject to inherent 384

uncertainties, due to the existence of other HCHO sources, in particular biomass burning 385

(during the dry season) and methane oxidation. Since these latter contributions are 386

favored by high temperature and radiation levels, they could possibly contribute to the 387

high correlation found between satellite-based isoprene and meteorological variabales. 388

For the ground-based emission, isoprene fluxes were determined by REA 389

measurements that were carried out for six days per month. Therefore, the low correlation 390


18

between ground-based isoprene fluxes and air temperature and PAR could partially result 391

from limited qualified data.392

Another factor correlated to ground-based isoprene fluxes is the leaf phenology 393

(in this study, LAI fractionated into age classes). The variation of mature LAI correlated 394

better to ground-based isoprene fluxes than to other factors (K34 site – R2=59%, p<0.05), 395

suggesting that the increasing isoprene emissions could partially follow the increasing of 396

mature leaves (Fig. 4). Wu et al. (2016) suggested that leaf demography (canopy leaf age 397

composition) and leaf ontogeny (age-dependent photosynthetic efficiency) are the main 398

reasons for the seasonal variation of the ecosystem photosynthetic capacity in Amazonia. 399

Since photosynthesis supplies the carbon to the methyl erythritol phosphate pathway to 400

produce isoprene (Delwiche and Sharkey, 1993; Harley et al., 1999; Lichtenthaler et al., 401

1997; Loreto and Sharkey, 1993; Rohmer, 2008; Schwender et al., 1997), and as isoprene 402

emissions are strongly dependent on leaf age and mainly emitted by mature leaves (Alves 403

et al., 2014), seasonal changes in the forest leaf-age fractions may also influence the 404

seasonality of isoprene emissions, suggesting higher emissions in the presence of more 405

mature leaves and during high ecosystem photosynthetic capacity efficiency.406

Understanding the correlations among light, temperature, leaf phenology (LAI 407

fractionated into age classes), and isoprene is not straightforward. The weak correlation 408

of seasonal changes between isoprene and light and temperature might be due to seasonal 409

changes in the isoprene dependency to environmental factors and biological factors. Light 410

and temperature peaked at the dry season; mature LAI, Gross Primary Productivity (GPP) 411

and photosynthetic capacity peaked at the wet season (Wu et al., 2016); and ground-412

based isoprene fluxes were high from the end of the dry to the dry-to-wet transition 413


19

seasons. This might suggest that isoprene emissions are stimulated by light and high 414

temperature during the beginning of the dry season and offset by the lower amount of 415

mature leaves. During the wet season, isoprene emissions could be stimulated by the 416

higher abundance of mature leaves and offset by the lower light availability and lower 417

temperature. But, at the end of the dry and at dry-to-wet transition seasons, there is a 418

combination of high light and high temperature with high amount of mature leaves, 419

possibly favoring high isoprene emissions. 420

This is supported by findings of a temperate plant species showing that LAI 421

dependency (changes in leaf age) was the most important factor affecting isoprene 422

emission capacity, but when LAI decreased, and senescence started at the end of the 423

summer, the isoprene dependency to PAR and air temperature was as high as the period 424

when PAR and air temperature reached their maximum (Brilli et al., 2016). This shows 425

seasonal variation in the strength of dependency to each factor that affects emissions. 426

Furthermore, we demonstrate a lack of a general correlation between ecosystem 427

seasonal cycles of photosynthetic capacity or GPP and isoprene emissions (Table 2). This 428

is consistent with previous studies that provide evidence that alternative non-429

photosynthetic pathways may contribute to isoprene synthesis under stress (Loreto and 430

Delfine, 2000), which may then lead to a decoupling of isoprene emission from 431

photosynthesis at high temperatures (Foster et al., 2014). In this light, it could be 432

suggested that the strong correlation between GPP and isoprene emission during leaf 433

phenology (Kuhn et al., 2004a) is reduced during conditions of high temperature. 434

As discussed above, separating the effects of changing temperature and light from 435

leaf phenology in canopy isoprene fluxes could allow for a more accurate quantification 436


20

and for a better understanding of seasonal isoprene flux. Here, we indicate that leaf 437

phenology plays an important role in seasonal variation of isoprene emissions, especially 438

because different leaf ages present different isoprene emission capacity and the 439

proportion of leaf age changes seasonally in Amazonia. However, when air temperature 440

is the highest, isoprene emission could be more stimulated by this factor, even though 441

mature LAI is still not at its maximum. We suggest future research to verify whether tree 442

species that present a regular seasonal leaf flushing are isoprene emitters and the strength 443

of those emissions by leaf age. 444

445

4.2. How can a consideration of leaf phenology observed in the field help to improve 446

model estimates of seasonal isoprene emissions? 447

Modeling of isoprene emissions from the Amazonian rainforest has been carried 448

out for around thirty years. The first models were simplified and parameterized with 449

observations from a few short field campaigns (see Table 1 of Alves et al., 2016). With 450

the increase in available data, more driving forces of isoprene emission were accounted 451

for in the latest versions of models, as the case of the MEGAN 2.1, which has been 452

improved with a multi-layer canopy model that accounts for light interception and leaf 453

temperature within the canopy, and includes changes in emissions due to leaf age that are 454

typically driven by satellite retrievals of LAI development (Guenther et al., 2012). 455

Results presented here are from MEGAN 2.1 estimates with local observations of 456

PAR, air temperature, and satellite-based leaf phenology. Initially, the default MEGAN 457

2.1 simulations did not fully capture the seasonal pattern of observed isoprene emission, 458

with none-significant correlation between model estimates and observations (R2= 0.16, 459


21

P>0.05, Table 2). This could be due to the near saturation of LAI seasonality in 460

Amazonian evergreen forests and poor representation of leaf age effect on isoprene 461

emission capacity of tropical tree species in the default MEGAN 2.1. Further, by using 462

the camera-derived LAI phenology and the leaf age demographics to update the leaf age 463

algorithm of the default MEGAN 2.1, we improved estimates of the proportion of leaves 464

in different leaf age categories for the site, but there were a lack of observations for 465

assigning the relative isoprene emission capacity for each age class. 466

It has been suggested that MEGAN uncertainties are mostly related to short-term 467

and long-term seasonality of the isoprene emission capacity (Niinemets et al., 2010). For 468

instance, for an Asian tropical forest, isoprene emission capacity was reported to be four 469

times lower than the default value of the MEGAN model (Langford et al., 2010), whereas 470

aircraft flux measurements in the Amazon were 35% higher than the MEGAN values (Gu 471

et al., 2017); and satellite retrievals suggested significantly lower isoprene emissions (30-472

40 % in Amazonia and northern Africa) with respect to the MEGAN-MOHYCAN 473

database (Bauwens et al., 2016). These all demonstrate that isoprene emission capacity is 474

not well represented in the model for regions where there are few or no measurements.475

With a sensitivity test, we parameterized the MEGAN 2.1 leaf age algorithm with 476

observed isoprene emission capacity among different leaf ages of E. coriacea (Alves et 477

al., 2014). The resulting simulation showed that by knowing the leaf age class 478

distribution and the isoprene emission capacity for each age class, MEGAN 2.1 estimates 479

can be improved and better agree with observations in terms of seasonal behavior. To 480

date, there is very little information about isoprene emission capacity for different leaf 481

ages of Amazonian plant species (Alves et al., 2014; Kuhn et al., 2004a). The scarcity of 482


22

observational studies in the field, along with the huge biodiversity and heterogeneity of 483

the Amazonian ecosystems, creates a challenge to optimize the isoprene emission 484

capacity parameterization in MEGAN and other models. Therefore, while introducing 485

local seasonal changes of canopy leaf age fractions in the model should improve 486

estimates, seasonal variations in isoprene emission capacity also need to be characterized 487

to better represent the effects of leaf phenology on ecosystem isoprene emissions. 488

489

4.3. Possible limitations 490

This study correlates available data of different scales and approaches. Thus, there 491

are limitations that need to be considered. One is the uncertainty related to the method 492

used to measure ground-based isoprene fluxes. The uncertainties of the REA flux 493

measurements ranged from 27.1% to 44.9% (more details in section 1 of Supplementary 494

Information). However, this study shows the largest dataset of seasonal isoprene fluxes in 495

Amazonia presented to date and results presented here are similar to previous 496

investigations, when same seasons are compared (see Table 1 of Alves et al., 2016). 497

Another limitation is the uncertainty of MEGAN estimates. It has been shown that 498

models tend to agree with observations within ~30% for canopy scale studies with site-499

specific parameters (Lamb et al., 1996). Here, part of the weak correlation between 500

observations and MEGAN 2.1 estimates is possibly due to short periods of measurements 501

and data gaps. There were data gaps of PAR and temperature for a few months in 2013. 502

This could influence the mean flux obtained from model estimates. Also, REA 503

measurements were carried out in intensive campaigns of six days per month, which may 504


23

not represent the flux for the entire month. Therefore, the limited data availability is still 505

challenging our understanding of isoprene emission seasonality. 506

507

5. Summary and Conclusions 508

To understand the pattern of isoprene seasonal fluxes in Amazonia is a difficult 509

task when considering the important role of Amazonian forests in accounting for global 510

BVOC and very limited field based observations in Amazonia. Seasonal variation of light 511

and temperature are thought to primarily drive isoprene seasonal emissions. However, 512

less notable factors might also influence ecosystem isoprene emission. Here, we suggest 513

that leaf phenology, especially when accounting for the effect of leaf demography 514

(canopy leaf age composition) and leaf ontogeny (age-dependent isoprene emission 515

capacity), has an important effect on seasonal changes of the ecosystem isoprene 516

emissions, which could play even more important role in regulating ecosystem isoprene 517

fluxes than light and temperature at seasonal timescale. 518

Albeit there are uncertainties related to measurements and modeling, results 519

presented here suggested that the unknown isoprene emission capacity for the different 520

leaf age classes found in the forest may be the main reason why MEGAN 2.1 did not 521

represent well the observed seasonality of isoprene fluxes. Additionally, part of these 522

model uncertainties arises because of a lack of representations of canopy structure and 523

light interception, including within-canopy variation in leaf functional traits; the leaf 524

phenology within the canopy; the physical processes by which isoprene is transported 525

within and above the forest canopy; chemical reactions that can take place within the 526

canopy; and, the most difficult to assess, emission variation due to the huge biodiversity 527


24

in Amazonia. Therefore, more detailed measurements of source and sink processes are 528

encouraged to improve our understanding of the seasonality of isoprene emissions in 529

Amazonia, which will improve surface emission models and will subsequently lead to a 530

better predictive vision of atmospheric chemistry, biogeochemical cycles, and climate. 531

5326. Data Availability 533

Even though the data are still not available in any public repository, the data are 534

available upon request from the main author. 535

536

7. Acknowledgements 537

The authors thank the National Institute for Amazonian Research (INPA) for continuous 538

support. We acknowledge the support by the Large Program of Biosphere-Atmosphere 539

Interactions (LBA) for the logistics and the micrometeorological group for their 540

collaboration concerning the meteorological parameters. We acknowledge Kolby Jardine 541

for providing the gas standard to calibrate the analytical system, and Paula Regina Corain 542

Lopes for the help in the fieldwork. J.W. is supported by DOE BER funded NGEE-543

Tropics project (contract # DE- SC00112704) to Brookhaven National Laboratory. 544

545

8. References 546

Adams, D. K., Fernandes, R. M. S., Holub, K. L., Gutman, S. I., Barbosa, H. M. J., 547Machado, L. A. T., Calheiros, A. J. P., Bennett, R. A., Robert Kursinski, E., Sapucci, L. 548F., Demets, C., Chagas, G. F. B., Arellano, A., Filizola, N., Rocha, A. A. A., Silva, R. A., 549Assunção, L. M. F., Cirino, G. G., Pauliquevis, T., Portela, B. T. T., Sá, A., De Sousa, J. 550M. and Tanaka, L. M. S.: The Amazon dense GNSS meteorological network: a new 551approach for examining water vapor and deep convection interactions in the Tropics, 552Bull. Amer. Meteor. Soc., 96(12), 2151–2165, doi:10.1175/BAMS-D-13-00171.1, 2015. 553 554


25

Adams, D. K., Fernandes, R. M. S. and Maia, J. M. F.: GNSS Precipitable Water Vapor 555from an Amazonian Rain Forest Flux Tower, J. Atmos. Ocean. Technol., 28(10), 1192–5561198, doi:10.1175/JTECH-D-11-00082.1, 2011. 557 558Adams, D. K., Gutman, S. I., Holub, K. L. and Pereira, D. S.: GNSS observations of deep 559convective time scales in the Amazon, Geophys. Res. Lett., 40(11), 2818–2823, 2013. 560 561Alves, E. G., Harley, P., Gonçalves, J. F. C., Moura, C. E. S. and Jardine, K.: Effects of 562light and temperature on isoprene emission at different leaf developmental stages of 563eschweilera coriacea in central amazon, Acta Amaz., 44(1), 9–18, doi:10.1590/S0044-56459672014000100002, 2014. 565 566Alves, E. G., Jardine, K., Tota, J., Jardine, A., Yãnez-Serrano, A. M., Karl, T., Tavares, 567J., Nelson, B., Gu, D., Stavrakou, T., Martin, S., Artaxo, P., Manzi, A. and Guenther, A.: 568Seasonality of isoprenoid emissions from a primary rainforest in central Amazonia, 569Atmos. Chem. Phys., 16(6), 3903–3925, doi:10.5194/acp-16-3903-2016, 2016. 570 571Andreae, M. O., Artaxo, P., Brandão, C., Carswell, F. E., Ciccioli, P., Da Costa, a. L., 572Gulf, a. D., Esteves, J. L., Gash, J. H. C., Grace, J., Kabat, P., Lelieveld, J., Malhi, Y., 573Manzi, a. O., Meixner, F. X., Nobre, a. D., Nobre, C., Ruivo, M. D. L. P., Silva-Dias, 574M. a., Stefani, P., Valentini, R., Von Jouanne, J. and Waterloo, M. J.: Biogeochemical 575cycling of carbon, water, energy, trace gases, and aerosols in Amazonia: The LBA-576EUSTACH experiments, J. Geophys. Res. D Atmos., 107(20), 577doi:10.1029/2001JD000524, 2002. 578 579Arneth, A., Monson, R. K., Schurgers, G., Niinemets, U. and Palmer, P. I.: Why are 580estimates of global terrestrial isoprene emissions so similar (and why is this not so for 581monoterpenes)? Atmos. Chem. Phys., 8(16), 4605–4620, doi:10.5194/acp-8-4605-2008, 5822008. 583 584Artaxo, P., Rizzo, L. V, Brito, J. F., Barbosa, H. M. J., Arana, A., Sena, E. T., Cirino, G. 585G., Bastos, W., Martin, S. T. and Andreae, M. O.: Atmospheric aerosols in Amazonia and 586land use change: from natural biogenic to biomass burning conditions, Faraday Discuss., 587165, 203, doi:10.1039/c3fd00052d, 2013. 588 589Barkley, M. P., Palmer, P. I., Kuhn, U., Kesselmeier, J., Chance, K., Kurosu, T. P., 590Martin, R. V., Helmig, D. and Guenther, A.: Net ecosystem fluxes of isoprene over 591tropical South America inferred from Global Ozone Monitoring Experiment (GOME) 592observations of HCHO columns, J. Geophys. Res., 113(D20), D20304, 593doi:10.1029/2008JD009863, 2008. 594 595Barkley, M. P., Palmer, P. I., De Smedt, I., Karl, T., Guenther, A. and Van Roozendael, 596M.: Regulated large-scale annual shutdown of Amazonian isoprene emissions? Geophys. 597Res. Lett., 36(4), L04803, doi:10.1029/2008GL036843, 2009. 598 599


26

Barkley, M. P., Smedt, I. De, Van Roozendael, M., Kurosu, T. P., Chance, K., Arneth, A., 600Hagberg, D., Guenther, A., Paulot, F., Marais, E. and Mao, J.: Top-down isoprene 601emissions over tropical South America inferred from SCIAMACHY and OMI 602formaldehyde columns, J. Geophys. Res. Atmos., 118(12), 6849–6868, 603doi:10.1002/jgrd.50552, 2013. 604 605Bauwens, M., Stavrakou, T., Müller, J. F., De Smedt, I., Van Roozendael, M., van der 606Werf, G. R., Wiedinmyer, C., Kaiser, J. W., Sindelarova, K. and Guenther, A.: Nine 607years of global hydrocarbon emissions based on source inversion of OMI formaldehyde 608observations, Atmos. Chem. Phys., 16(15), 10133–10158, doi:10.5194/acp-16-10133-6092016, 2016. 610 611Bauwens, M., Stavrakou, T., Müller, J.-F., Van Schaeybroeck, B., De Cruz, L., De Troch, 612R., Giot, O., Hamdi, R., Termonia, P., Laffineur, Q., Amelynck, C., Schoon, N., 613Heinesch, B., Holst, T., Arneth, A., Ceulemans, R., Sanchez-Lorenzo, A., and Guenther, 614A.: Recent past (1979–2014) and future (2070–2099) isoprene fluxes over Europe 615simulated with the MEGAN-MOHYCAN model, Biogeosciences Discuss., 616https://doi.org/10.5194/bg-2017-532, in review, 2017. 617 618Bowling, D. R., Turnipseed, A. A., Delany, A. C., Baldocchi, D. D., Greenberg, J. P. 619and Monson, R. K.: The use of relaxed eddy accumulation to measure biosphere-620atmosphere exchange of isoprene and other biological trace gases, Oecologia, 116(3), 621306–315, doi:10.1007/s004420050592, 1998. 622 623Brilli, F., Gioli, B., Fares, S., Terenzio, Z., Zona, D., Gielen, B., Loreto, F., Janssens, I. 624A. and Ceulemans, R.: Rapid leaf development drives the seasonal pattern of volatile 625organic compound (VOC) fluxes in a “coppiced” bioenergy poplar plantation, Plant, Cell 626Environ., 39(3), 539–555, doi:10.1111/pce.12638, 2016. 627 628Claeys, M., Graham, B., Vas, G., Wang, W., Vermeylen, R., Pashynska, V., Cafmeyer, J., 629Maenhaut, W., Guyon, P., Andreae, M. O. and Artaxo, P.: Formation of secondary 630organic aerosols through photooxidation of isoprene, Science, 303(5661), 1173–1176, 631doi:10.1126/science.1092805, 2004. 632 633de Araújo, A. C., Dolman, A. J., Waterloo, M. J., Gash, J. H. C., Kruijt, B., Zanchi, F. 634B., de Lange, J. M. E., Stoevelaar, R., Manzi, A. O. and Nobre, A. D.: The spatial 635variability of CO2 storage and the interpretation of eddy covariance fluxes in central 636Amazonia, Agric. For. Meteorol., 150(2), 226–237, doi:10.1016/j.agrformet.2009.11.005, 6372010. 638 639Delwiche, C. F. and Sharkey, T. D.: Rapid appearance of 13C in biogenic isoprene when 64013CO2 is fed to intact leaves, Plant. Cell Environ., 16(5), 587–591, doi:10.1111/j.1365-6413040.1993.tb00907.x, 1993. 642 643Foster, P. N., Prentice, I. C., Morfopoulos, C., Siddall, M. and Van Weele, M.: Isoprene 644emissions track the seasonal cycle of canopy temperature, not primary production: 645


27

Evidence from remote sensing, Biogeosciences, 11(13), 3437–3451, doi:10.5194/bg-11-6463437-2014, 2014. 647 648Goldstein, A. H. and Galbally, I. E.: Known and Unexplored Organic Constituents in the 649Earth’s Atmosphere, Environ. Sci. Technol., 41(5), 1514–1521, doi:10.1021/es072476p, 6502007. 651 652Gu, D., Guenther, A. B., Shilling, J. E., Yu, H., Huang, M., Zhao, C., Yang, Q., Martin, 653S. T., Artaxo, P., Kim, S., Seco, R., Stavrakou, T., Longo, K. M., Tóta, J., de Souza, R. 654A. F., Vega, O., Liu, Y., Shrivastava, M., Alves, E. G., Santos, F. C., Leng, G. and Hu, 655Z.: Airborne observations reveal elevational gradient in tropical forest isoprene 656emissions, Nat. Commun., 8-, 15541, doi:10.1038/ncomms15541, 2017. 657 658Guenther, A. B., Jiang, X., Heald, C. L., Sakulyanontvittaya, T., Duhl, T., Emmons, L. 659K. and Wang, X.: The Model of Emissions of Gases and Aerosols from Nature version 6602.1 (MEGAN2.1): an extended and updated framework for modeling biogenic emissions, 661Geosci. Model Dev., 5(2), 1503–1560, doi:10.5194/gmdd-5-1503-2012, 2012. 662 663Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P. I. and Geron, C.: 664Estimates of global terrestrial isoprene emissions using MEGAN (Model of Emissions of 665Gases and Aerosols from Nature), Atmos. Chem. Phys., 6(1), 3181–3210, 666doi:10.5194/acpd-6-107-2006, 2006. 667 668Harley, P., Monson, R. and Lerdau, M.: Ecological and evolutionary aspects of isoprene 669emission from plants, Oecologia, 118, 109–123, 1999. 670 671Harley, P., Vasconcellos, P., Vierling, L., Pinheiro, C. C. D. S., Greenberg, J., Guenther, 672A., Klinger, L., Almeida, S. S. De, Neill, D., Baker, T., Phillips, O., Malhi, Y. and De 673Almeida, S. S.: Variation in potential for isoprene emissions among Neotropical forest 674sites, Glob. Chang. Biol., 10(5), 630–650, doi:10.1111/j.1529-8817.2003.00760.x, 2004. 675 676Jardine, K., Chambers, J., Alves, E. G., Teixeira, A., Garcia, S., Holm, J., Higuchi, N., 677Manzi, A., Abrell, L., Fuentes, J. D., Nielsen, L. K., Torn, M. S. and Vickers, C. E.: 678Dynamic balancing of isoprene carbon sources reflects photosynthetic and 679photorespiratory responses to temperature stress, Plant Physiol., 166(4), 2051–2064, 680doi:10.1104/pp.114.247494, 2014. 681 682Karl, T., Guenther, A., Yokelson, R. J., Greenberg, J., Potosnak, M., Blake, D. R. and 683Artaxo, P.: The tropical forest and fire emissions experiment: Emission, chemistry, and 684transport of biogenic volatile organic compounds in the lower atmosphere over 685Amazonia, J. Geophys. Res., 112(D18), D18302, doi:10.1029/2007JD008539, 2007. 686 687Karl, T., Potosnak, M., Guenther, A. B., Clark, D., Walker, J., Herrick, J. D. and Geron, 688C.: Exchange processes of volatile organic compounds above a tropical rain forest: 689Implications for modeling tropospheric chemistry above dense vegetation, J. Geophys. 690Res., 109(D18), D18306, doi:10.1029/2004JD004738, 2004. 691


28

Kesselmeier, J., Ciccioli, P., Kuhn, U., Stefani, P., Biesenthal, T., Rottenberger, S., Wolf, 692A., Vitullo, M., Valentini, R., Nobre, A., Kabat, P. and Andreae, M. O.: Volatile organic 693compound emissions in relation to plant carbon fixation and the terrestrial carbon budget, 694Global Biogeochem. Cycles, 16(4), 73-1-73–9, doi:10.1029/2001GB001813, 2002. 695 696Kuhn, U., Rottenberger, S., Biesenthal, T., Wolf, A., Schebeske, G., Ciccioli, P., 697Brancaleoni, E., Frattoni, M., Tavares, T. M. and Kesselmeier, J.: Isoprene and 698monoterpene emissions of Amazonian tree species during the wet season: Direct and 699indirect investigations on controlling environmental functions, J. Geophys. Res., 700107(D20), 8071, doi:8071 10.1029/2001jd000978, 2002. 701 702Kuhn, U., Rottenberger, S., Biesenthal, T., Wolf, A., Schebeske, G., Ciccioli, P. and 703Kesselmeier, J.: Strong correlation between isoprene emission and gross photosynthetic 704capacity during leaf phenology of the tropical tree species Hymenaea courbaril with 705fundamental changes in volatile organic compounds emission composition during early 706leaf development, Plant, Cell Environ., 27(12), 1469–1485, doi:10.1111/j.1365-7073040.2004.01252.x, 2004a. 708 709Kuhn, U., Rottenberger, S., Biesenthal, T., Wolf, a., Schebeske, G., Ciccioli, P., 710Brancaleoni, E., Frattoni, M., Tavares, T. M. and Kesselmeier, J.: Seasonal differences in 711isoprene and light-dependent monoterpene emission by Amazonian tree species, Glob. 712Chang. Biol., 10(5), 663–682, doi:10.1111/j.1529-8817.2003.00771.x, 2004b. 713 714Lamb, B., Pierce, T., Baldocchi, D., Allwine, E., Dilts, S., Westberg, H., Geron, C., 715Guenther, A., Klinger, L., Harley, P. and Zimmerman, P.: Evaluation of forest canopy 716models for estimating isoprene emissions, J. Geophys. Res., 101(D17), 22787–22797, 717doi:10.1029/96JD00056, 1996. 718 719Langford, B., Misztal, P. K., Nemitz, E., Davison, B., Helfter, C., Pugh, T. A. M., 720MacKenzie, A. R., Lim, S. F. and Hewitt, C. N.: Fluxes and concentrations of volatile 721organic compounds from a South-East Asian tropical rainforest, Atmos. Chem. Phys., 72210(17), 8391–8412, doi:10.5194/acp-10-8391-2010, 2010. 723 724Lichtenthaler, H. K., Rohmer, M. and Schwender, J.: Two independent biochemical 725pathways for isopentenyl diphosphate and isoprenoid biosynthesis in higher plants, 726Physiol. Plant., 101(3), 643–652, doi:10.1111/j.1399-3054.1997.tb01049.x, 1997. 727 728Lopes, A. P., Nelson, B. W., Wu, J., Graça, P. M. L. de A., Tavares, J. V., Prohaska, N., 729Martins, G. A. and Saleska, S. R.: Leaf flush drives dry season green-up of the Central 730Amazon, Remote Sens. Environ., 182, 90–98, doi:10.1016/j.rse.2016.05.009, 2016. 731 732Loreto, F. and Delfine, S.: Emission of Isoprene from Salt-stressed Eucalyptus globulus 733Leaves., Plant Physiol., 123(4), 1605–1610, doi:10.1104/pp.123.4.1605, 2000. 734 735


29

Loreto, F. and Sharkey, T. D.: On the relationship between isoprene emission and 736photosynthetic metabolites under different environmental conditions, Planta, 189(3), 737420–424, doi:10.1007/BF00194440, 1993. 738 739Machado, L. A. T., Laurent, H., Dessay, N. and Miranda, I.: Seasonal and diurnal 740variability of convection over the Amazonia: A comparison of different vegetation types 741and large scale forcing, Theor. Appl. Climatol., 78(1–3), 61–77, doi:10.1007/s00704-742004-0044-9, 2004. 743 744Myneni, R.: MCD15A2H MODIS/Terra+Aqua Leaf Area Index/FPAR 8-day L4 745Global 500m SIN Grid V006. NASA EOSDIS Land Processes DAAC, 746https://doi.org/10.5067/modis/mcd15a2h.006, 2015.747 748Nelson, B., Tavares, J. V., Wu, J., Lopes, A. P., Marostica, S., Martins, G., Prohaska, N., 749Albert, L. P., de Araújo, A. C., Manzi, A. O., Saleska, S., Huete, A.: Seasonality of 750central Amazon Forest Leaf Flush using tower mounted RGB Camera. AGU Fall 751Meeting, San Francisco, California, 15–19 December 2014, B11G-0107, 2014. 752 753Niinemets, U., Arneth, A., Kuhn, U., Monson, R. K., Penuelas, J. and Staudt, M.: The 754emission factor of volatile isoprenoids: stress, acclimation, and developmental responses, 755Biogeosciences, 7(7), 2203–2223, doi:10.5194/bg-7-2203-2010, 2010. 756 757Rennó, C. D., Nobre, A. D., Cuartas, L. A., Soares, J. V., Hodnett, M. G., Tomasella, J. 758and Waterloo, M. J.: HAND, a new terrain descriptor using SRTM-DEM: Mapping terra-759firme rainforest environments in Amazonia, Remote Sens. Environ., 112(9), 3469–3481, 760doi:10.1016/j.rse.2008.03.018, 2008. 761 762Rohmer, M.: From molecular fossils of bacterial hopanoids to the formation of isoprene 763units: Discovery and elucidation of the methylerythritol phosphate pathway, Lipids, 76443(12), 1095–1107, doi:10.1007/s11745-008-3261-7, 2008. 765 766Schwender, J., Zeidler, J., Gröner, R., Müller, C., Focke, M., Braun, S., Lichtenthaler, F. 767W. and Lichtenthaler, H. K.: Incorporation of 1-deoxy-D-xylulose into isoprene and 768phytol by higher plants and algae, FEBS Lett., 414(1), 129–134, doi:10.1016/S0014-7695793(97)01002-8, 1997. 770 771Stavrakou, T., Müller, J.-F., Bauwens, M., De Smedt, I., Van Roozendael, M., Guenther, 772A., Wild, M. and Xia, X.: Isoprene emissions over Asia 1979–2012: impact of climate 773and land-use changes, Atmos. Chem. Phys., 14(9), 4587–4605, doi:10.5194/acp-14-4587-7742014, 2014. 775 776Stavrakou, T., Müller, J.-F., Bauwens, M., De Smedt, I., Van Roozendael, M., De 777Mazière, M., Vigouroux, C., Hendrick, F., George, M., Clerbaux, C., Coheur, P.-F. and 778Guenther, A.: How consistent are top-down hydrocarbon emissions based on 779formaldehyde observations from GOME-2 and OMI? Atmos. Chem. Phys., 15(20), 78011861–11884, doi:10.5194/acp-15-11861-2015, 2015. 781


30

782Stavrakou, T., Müller, J.-F., De Smedt, I., Van Roozendael, M., van der Werf, G. R., 783Giglio, L. and Guenther, A.: Global emissions of non-methane hydrocarbons deduced 784from SCIAMACHY formaldehyde columns through 2003–2006, Atmos. Chem. Phys., 9, 7853663-3679, https://doi.org/10.5194/acp-9-3663-2009, 2009. 786 787Tóta, J., Fitzjarrald, D. R. and da Silva Dias, M. A. F.: Amazon rainforest exchange of 788carbon and subcanopy air flow: Manaus LBA site – a complex terrain condition, 789Scientific World Journal., 2012, 165067, doi:10.1100/2012/165067, 2012. 790 791Wu, J., Albert, L. P., Lopes, A. P., Restrepo-Coupe, N., Hayek, M., Wiedemann, K. T., 792Guan, K., Stark, S. C., Christoffersen, B., Prohaska, N., Tavares, J. V., Marostica, S., 793Kobayashi, H., Ferreira, M. L., Campos, K. S., Silva, R. da, Brando, P. M., Dye, D. G., 794Huxman, T. E., Huete, A. R., Nelson, B. W. and Saleska, S. R.: Leaf development and 795demography explain photosynthetic seasonality in Amazon evergreen forests, Science, 796351(6276), doi:10.1126/science.aad5068, 2016. 797 798Yáñez-Serrano, A. M., Nölscher, a C., Williams, J., Wolff, S., Alves, E., Martins, G. a, 799Bourtsoukidis, E., Brito, J., Jardine, K. J., Artaxo, P. and Kesselmeier, J.: Diel and 800seasonal changes of biogenic volatile organic compounds within and above an 801Amazonian rainforest, Atmos. Chem. Phys., 15, 3359–3378, doi:10.5194/acp-15-3359-8022015, 2015. 803 804

Tables 805

Table 1: Environmental and biological factors used to input the MEGAN 2.1: number of 806days with data available for each variable for the year 2013 807

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec PAR n=31 n=28 n=31 n=30 n=31 n=30 n=31 n=15 n=30 n=18 n=19 n=15 Air temperature

n=31 n=28 n=31 n=30 n=31 n=30 n=31 n=15 n=30 n=18 n=19 n=15

CAMERA-LAI*

n=5 n=4 n=5 n=5 n=5 n=5 n=5 n=5 n=5 n=5 n=5 n=5

MODIS-LAI**

n=4 n=4 n=4 n=3 n=5 n=4 n=4 n=4 n=4 n=3 n=4 n=4

* Number of days with images analyzed to derive CAMERA-LAI as described in section 2.4. 808** Number of days that the satellite passed over the site domain. 809

810

811

812

813

814


31

Table 2: Correlation coefficient, R, of regressions for ground-based isoprene flux, 815satellite-derived isoprene flux, environmental factors, biological factors, and 816MEGAN 2.1 simulations 817 Ground-based

isoprene flux Satellite-derived

isoprene flux (2013 year)

PAR 0.007a 0.55c

PAR – REA measurement days 0.11a -----

Air temperature 0.15a 0.79c

Air temperature – REA measurement days 0.39a -----

young LAI 0.04a 0.35b

mature LAI 0.59b 0.05a

old LAI -0.6b -0.4b

Photosynthetic capacity* 0.49a -----

GPP* 0.36a -----

MEGAN (MODIS-LAI) 0.16a 0.76c

MEGAN (CAMERA-LAI) 0.11a 0.67c

MEGAN (MODIS-LAI) EAF changed 0.19a 0.66c

MEGAN (CAMERA-LAI) EAF changed 0.52b 0.59c

Ground-based isoprene flux ----- 0.13a

PAR,photosynthetic active radiation; GPP, gross primary productivity;818EAF, emission activity factor; 819* Data from Wu et al. (2016) 820a not statistically significant (P > 0.05) 821b statistically significant (P < 0.05) 822c statistically significant (P < 0.001) 823824

Figure captions 825

Figure 1. Location of the experimental site in central Amazonia – K34 tower. Hill-826

shaded digital elevation data used as background topography is from the Shuttle Radar 827

Topography Mission, with resolutions of ~900m (top panel) and ~30m (lower panel). 828

White ring indicates two km radius around the flux tower. Elevation scale for lower panel 829

is "meters above sea level". 830


32

Figure 2. Monthly averages of photosynthetic active radiation (PAR) (a) and air 831

temperature (b) from 2005 to 2013 at the K34 tower site (measured every 30 min during -832

6:00-18:00h, local time). OMI satellite-derived isoprene flux in a resolution of 0.5º 833

centered on K34 tower site from 2005 to 2013 (c). Monthly averages of isoprene flux 834

were scaled to 10:00-14:00, local time. Error bars represent one standard error of the 835

mean. 836

Figure 3. Monthly cumulative precipitation given by the Tropical Rainfall Measuring 837

Mission (TRMM) for the K34 tower domain in 2013 (a) Monthly averages of PAR (b) 838

and air temperature (c), both measured every 30 minutes during 6:00-18:00h, local time, 839

at the K34 tower site in 2013. Isoprene flux measured with the REA system at the K34 840

tower site in 2013 (d). 841

Figure 4. CAMERA-LAI derived for the K34 tower site. CAMERA-LAI data are 842

presented in three different leaf age classes: young LAI, mature LAI and old LAI. Error 843

bars represent one standard deviation from the mean. Background color shadings indicate 844

each season and are explicit in the legend. DWT season and WDT season stand for the 845

dry-to-wet transition season and the wet-to-dry transition season, respectively. 846

Figure 5: Isoprene flux observed (REA) and estimated with MEGAN 2.1 default mode, 847

leaf age algorithm driven by MODIS-LAI, and with MEGAN 2.1 leaf age algorithm 848

driven by CAMERA-LAI. EAF stands for emission activity factor, which was changed 849

for the different leaf age classes based on emissions of E. coriacea (Alves et al., 2014).850

Figure 6. Emission activity factor (EAF) of isoprene for each leaf age class assigned in 851

the default mode of MEGAN 2.1 proportional to leaf age class distribution derived from 852

field observations (CAMERA-LAI) (a) Isoprene EAF for each leaf age class, obtained 853


33

from leaf level measurements of the tree species E. coriacea, proportional to leaf age 854

class distribution derived from field observations (CAMERA-LAI) (b) Observations of 855

the tree species E. coriacea (Alves et al., 2014) and CAMERA-LAI are both from the 856

K34 site. 857

858

Figures 859

860Figure 1. Location of the experimental site in central Amazonia – K34 tower. Hill-861

shaded digital elevation data used as background topography is from the Shuttle Radar 862

Topography Mission, with resolutions of ~900m (top panel) and ~30m (lower panel). 863

White ring indicates two km radius around the flux tower. Elevation scale for lower panel 864

is "meters above sea level". 865


34

866Figure 2. Monthly averages of photosynthetic active radiation (PAR) (a) and air 867

temperature (b) from 2005 to 2013 at the K34 tower site (measured every 30 min during -868

6:00-18:00h, local time). OMI satellite-derived isoprene flux in a resolution of 0.5º 869

centered on K34 tower site from 2005 to 2013 (c). Monthly averages of isoprene flux 870

were scaled to 10:00-14:00, local time. Error bars represent one standard error of the 871

mean. 872

873


35

874Figure 3. Monthly cumulative precipitation given by the Tropical Rainfall Measuring 875

Mission (TRMM) for the K34 tower domain in 2013 (a) Monthly averages of PAR (b) 876

and air temperature (c), both measured every 30 minutes during 6:00-18:00h, local time, 877

at the K34 tower site in 2013. Isoprene flux measured with the REA system at the K34 878

tower site in 2013 (d). 879

880

881

882

883


36

884Figure 4. CAMERA-LAI derived for the K34 tower site. CAMERA-LAI data are 885

presented in three different leaf age classes: young LAI, mature LAI and old LAI. Error 886

bars represent one standard deviation from the mean. Background color shadings indicate 887

each season and are explicit in the legend. DWT season and WDT season stand for the 888

dry-to-wet transition season and the wet-to-dry transition season, respectively. 889

890Figure 5: Isoprene flux observed (REA) and estimated with MEGAN 2.1 default mode, 891

leaf age algorithm driven by MODIS-LAI, and with MEGAN 2.1 leaf age algorithm 892

driven by CAMERA-LAI. EAF stands for emission activity factor, which was changed 893

for the different leaf age classes based on emissions of E. coriacea (Alves et al., 2014).894


37

895Figure 6. Emission activity factor (EAF) of isoprene for each leaf age class assigned in 896

the default mode of MEGAN 2.1 proportional to leaf age class distribution derived from 897

field observations (CAMERA-LAI) (a) Isoprene EAF for each leaf age class, obtained 898

from leaf level measurements of the tree species E. coriacea, proportional to leaf age 899

class distribution derived from field observations (CAMERA-LAI) (b) Observations of 900

the tree species E. coriacea (Alves et al., 2014) and CAMERA-LAI are both from the 901

K34 site. 902

903904905


1 Leaf phenology as one important driver of seasonal ...

Documents

Transcript of 1 Leaf phenology as one important driver of seasonal ...